Some surprises about the seasonal flu
Pretty boring report this time, so I’ll talk about my little “flu project” today.
My goal today was to compare the so-called “Case Fatality Rate” of COVID with that of the seasonal flu. Well, it’s simply not possible. There is no national reporting system for the flu. More on that later…
First, I want to clear up confusion about Case Fatality Rate (CFR), Infection Fatality Rate (IFR), and Mortality Rate (the conditional probability of dying if you contract the disease). First, let me say that all of these terms are sometimes used interchangeably, and that adds to the confusion. Here is my best explanation from an actuarial viewpoint.
CASE FATALITY RATE (CFR): This is quoted as nothing more than the total number of recorded deaths divided by the number of confirmed cases. You can find it on sites like www.infection2020.com on the home page. It’s currently running about 6%. It is NOT a representation of the probability of dying if you have COVID, simply because the number of confirmed cases does not represent the number of people who have COVID in society – that number is much larger based on the several random testing studies done in the United States.
INFECTION FATALITY RATE (IFR): This is “usually” the number of recorded deaths divided by an estimate of the total number of people in society who have or have had COVID. This number is much lower, but rarely quoted. It is somewhat closer to a mortality rate, but until the disease has run its course, it will be less than the mortality rate, since deaths lag the case count.
MORTALITY RATE: Yes, this is the probability of dying if you catch COVID. The mortality rate is the number of recorded deaths divided by an estimated number of cases that the deaths originate from. It involves modeling the total COVID in society, and then matching that up with the deaths that result from that cohort of cases. This is more complicated to do, but that’s what actuaries are for (literally). My first (but not last) mortality calculation (described at the end of each day’s report) resulted in an overall mortality rate of 0.53%, but of course wildly different by age group.
Now, on to the flu. It turns out that no one actually counts flu cases or deaths – hey, I’m just a mathematician, so I was surprised by this. It’s easy to count flu tests and laboratory confirmed cases, but not deaths. I thought I would be clever and divide flu confirmed deaths from last year by positive flu tests and show that the CFR for flu is higher than COVID. However, the number that doesn’t exist is “confirmed flu deaths”.
[all flu data to follow from the CDC] Last season (2018/2019) we administered 1,454,484 flu tests in the U.S. 229,364 of these were positive – yes, that’s right – less than a quarter of a million confirmed flu cases. Yet, the CDC estimates that 36 million had the flu, and there were 34,000 deaths. How do they know this? Well, they don’t really. The confidence interval is very wide. They extrapolate from a survey of hospitals and medical providers in 13 geographical areas about respiratory illness visits to estimate the number of flu patients who sought medical care. Then they estimate the number that did not seek medical care from the 2010 Behavioral Risk Factor Surveillance Survey. From this, they model how many likely had the flu nationally. There is a similar methodology for deaths, since a minority of flu deaths show flu on the death certificate. The end result is that I couldn’t complete my little project, but you get the idea. Very likely more people died from the flu than the total number of confirmed cases… (if I were a reporter I’d probably conclude that the mortality rate for the seasonal flu was over 100%).
Sorry for the long diatribe, but sometimes when I look into things the answers floor me.
My next project is to work on a refined mortality model. My goal this time is to use all of the random testing studies to date, and gain better insight on the mortality rate by age and morbidity.
As always, feel free to send me your questions about my assumptions, methodology, or modeling in general.
- Likely date of active case peak (Chalke modeling): April 10
- Likely date of peak deaths (IHME): April 16 (last revision on May 20)
- Short term projection for active cases tomorrow: 155,000
- Total Test Results reported today: 408,415 (very high)
- Total Pending tests reported today: 3,641 (very low)
- National reported case Growth Rate today: 1.6% (very low)